Report of the Ad-hoc Group on the Impact of Sonar on Cetaceans and Fish (AGISC)

Transcription

1 ICES Advisory Committee on Ecosystems ICES CM 2005/ACE:01 Report of the Ad-hoc Group on the Impact of Sonar on Cetaceans and Fish (AGISC) By correspondence International Council for the Exploration of the Sea Conseil International pour l Exploration de la Mer H.C. Andersens Boulevard 44-46, 1553 Copenhagen V, Denmark

2 This report is not to be quoted without prior consultation with the General Secretary. The document is a report of an Expert Group under the auspices of the International Council for the Exploration of the Sea and does not necessarily represent the views of the Council. ii ICES 2004 AGISC Report

3 TABLE OF CONTENTS Section Page 1 INTRODUCTION Participation Terms of Reference Justification of Terms of Reference Framework for response Overview by the Chair Acknowledgements PHYSICAL BACKGROUND Units for measuring noise Use of the decibel scale in water Parameters for estimating noise Source level Impulsive sound Sound propagation and transmission loss Ambient noise Sonar in general Low-frequency sonar Mid frequency sonar High frequency sonar BIOLOGICAL BACKGROUND Hearing in cetaceans Anatomy and physiology Hearing in smaller odontocetes Hearing in larger odontocetes Hearing in mysticetes Potential effects of sound on cetaceans Direct damage to hearing Non-auditory tissue damage Masking and changes in vocal behaviour Behavioural reactions CETACEANS AND SONAR Marine mammals Beaked whales Review of literature on effects of sonar on beaked whales Case study: Greece Case study: Bahamas Case study: Canary Islands Other cetaceans and sonar Research on LFA and cetaceans SUMMARY OF GAPS IN UNDERSTANDING OTHER RELEVANT ITEMS Noise pollution as a more serious problem? GENERAL CONCLUSION RECOMMENDATIONS Future investigations and research Mitigation measures for cetaceans Introduction Control at source Mitigation of death and injury caused by the direct effects of sound Other control methods Monitoring Current mitigation measures REFERENCES ANNEX 1: NAMES AND ADDRESSES OF PARTICIPANTS WHO CONTRIBUTED TO THIS REPORT... # iii

4

5 1 INTRODUCTION 1.1 Participation The following members of the Ad hoc Group on the Impact of Sonar on Cetaceans and Fish (AGISC) participated in producing this report (see Annex 1 for addresses). Chris Clark Antonio Fernández Alexandros Frantzis Roger Gentry Jonathan Gordon Tony Hawkins Paul Jepson Finn Larsen Jeremy Nedwell Mark Tasker (chair) Jacob Tougaard Peter Tyack Tana Worcester USA Spain Greece USA UK UK UK Denmark UK UK Denmark USA Canada 1.2 Terms of Reference At the MCAP meeting January 2004, an Ad hoc Group on the Impact of Sonar on Cetaceans and Fish (AGISC) was established and was given the following terms of reference: Review and evaluate all relevant information concerning the impact of sonar on cetaceans and fish; Identify the gaps in our current understanding; Prepare recommendations for future investigations and research; Prepare draft advice on possible mitigation measures to reduce or minimize the impact of sonar on cetaceans and fish. 1.3 Justification of Terms of Reference The terms of reference derive from a letter from Catherine Day (Director General of EC DG Environment) to David Griffith (General Secretary, ICES), dated 25 September In this letter, the European Commission indicated that it had for some time received complaints about the impact of sonar on marine mammals. These complaints claimed that the emission of intense, low and medium frequency tone bursts has a disturbing effect on cetaceans. Information had also been forwarded indicating that these sonars might have an impact on fish and fish behaviour. European legislation (mainly the Habitats Directive (92/43/EC)) requires Member States of the European Union to take measures to establish a system of strict protection for all cetaceans in European waters. The European Commission does not have a comprehensive and authoritative review of information concerning the impact of sonar, and thus finds it difficult to develop a clear position on the issue. The Commission therefore asked ICES to undertake a scientific review and evaluation of relevant information concerning the impact of sonar on cetaceans and fish, to identify the gaps in current understanding and to make recommendations for future investigations or research. The Commission is also interested in advice on possible mitigation measures to reduce or minimise the impact of sonar on cetaceans and fish. 1.4 Framework for response This is the first of two reports. It deals with the physical background on the issue and with the relevant biology and response of cetaceans. The second report will cover the relevant biology and response of fish. The Group s response to these terms of reference has been compiled by correspondence. Sections were initially drafted by Group Members and then agreed by circulation to all members. Much of the report is based on existing review literature (not all relating to sonar directly), updated and amended as appropriate. 5 ICES 2004 AGISC Report

6 1.5 Overview by the Chair The effects of human inputs to natural systems have been a topic of interest and study for many years, however much the greatest amount of work has been carried out on chemical inputs, both in the form of contaminants and nutrients. The subject of energy input has historically received much less attention. The anthropogenic input of sound to the marine environment started with the coming of mechanically propelled ships, but until the advent of sonar, nearly all sound input was a by-product of another activity as opposed to deliberate. Both forms of input though carry the risk of affecting other marine life. Evidence has been available for some time that anthropogenic noise has the capacity to disturb those forms of marine life dependent on sound for communication and sensing in the seas. Much less evidence has been available on damage, injury or lethal consequences at the individual level, and none at the population level. A series of incidents in recent years when certain deep-diving whale species stranded or died co-incident with the use of high-powered sonar alerted many more to the risks posed by sound. Research elsewhere indicated that other forms of loud sound might affect fish. This first report of the group sprang from concerns about the effects of sonar on cetaceans. The behaviour of sound in the marine environment is complex and is equally complex to describe. We have attempted to describe the physical background briefly in the first main part of this report, but are aware that this may be too brief and simplistic for some. We refer those readers to standard texts for further information. This section includes a brief description of the types of sonar in use today. It proved difficult to find information on the characteristics of many forms of sonar. The next section of the report deals with the mechanisms for hearing in cetaceans and describes the potential effects of sound on these mechanisms and the behaviour of these animals. Until recently, most concern has focussed on the effects on hearing and communication systems of cetaceans but recent evidence is indicating that damage may also be caused through other mechanisms, and perhaps indirectly through dangerous alterations in behaviour. There is however little experimental evidence currently in the public domain of the effects of sonar on the acoustic systems, physiology or behaviour of cetaceans. Logistically, any such experiments are easiest to conduct in the laboratory on individual animals, but extrapolating the few available items of information to wild populations is at present very difficult and uncertain. Section 4 reviews observations and deductions from cases where whales have stranded or been found dead in association with the nearby use of military sonar. As with many observational cases, obtaining the best and most pertinent evidence proved difficult both from the corpses and in other cases of strandings, from the military authorities. In order to improve deductions, we ideally need to complete three of four cells in a 2 x 2 matrix - naval operations occurred or not occurred versus marine mammal strandings occurred or not occurred. We have knowledge of some stranding events associated with naval operations, and possibly some information on the number of stranding events without naval sonar being present, but we do not know how many naval sonar operations occurred (in suitable beaked whale habitat) without any observed marine mammal strandings. It is though agreed that high-powered, mid-frequency sonar can affect beaked whales in particular. These effects can lead to death, either at sea or as a consequence of stranding ashore. These effects may be caused by a lethal behavioural change leading to physiological damage, or possibly by direct physiological damage. Hypotheses exist to explain these effects. It seems likely that these effects also occur at lower received sound levels than previously thought likely to cause damage and as a consequence the sphere of effect of these sonars is not known. Coupled with the lack of knowledge of the population size or distribution of beaked whales, we cannot be certain of whether population level effects might occur. However, at present it appears that these military sonars are not used widely. This could change in the future if these sonars were more widely deployed on ships or were used in non-exercise situations. Effects would be most severe in areas important for beaked whales. Section 5 outlines some of the gaps in understanding around this issue and makes some suggestions as to how they might be addressed. Section 6 notes that other facets of the issue of noise in the ocean could have potentially more significant effects than direct lethal effects on individuals. In particular, the apparently increasing levels of anthropogenic low-frequency noise (mostly from shipping) may have consequences for the large baleen whales that use these frequencies for communication. Section 8 provides the groups recommendations that may form the basis of advice from ICES. 1.6 Acknowledgements We thank Rene Swift and Jay Barlow for help in accessing some of the references used in this report. Gerald D'Spain, Jim Miller, and Dave Bradley provided comments on noise budgets. Bertel Møhl and Hans Lassen both provided helpful comments. Jake Rice read the whole report and provided many helpful comments. ICES 2004 AGISC Report 6

7 2 PHYSICAL BACKGROUND 2.1 Units for measuring noise Underwater sound is usually expressed using the logarithmic decibel (decibel) scale. Underwater sound is conventionally presented in decibels referenced to 1 micropascal, i.e. as db re 1 µpa, and this convention has been adhered to in this report Use of the decibel scale in water The fundamental unit of sound pressure is the Newton per square metre, or Pascal. However, when describing underwater acoustic phenomena it is normal to express the sound pressure through the use of a logarithmic scale termed the Sound Pressure Level. There are two reasons for this. First, there is a very wide range of sound pressures measured underwater, from around Pa in quiet sea to around 10,000,000 Pa for an explosive blast. The use of a logarithmic scale compresses the range so that it can be easily described (in this example, from 0 db to 260 db re 1 µpa). Second, many of the mechanisms affecting sound underwater cause loss of sound at a constant rate when it is expressed on the db scale. The Sound Pressure Level (SPL) is defined as: SPL = 20 log (P/P ref ) where P is the sound pressure to be expressed on the scale and P ref is the reference pressure, which for underwater applications is 1 µpa. For instance, a pressure of 1 Pa would be expressed as an SPL of 120 db re 1 µpa. 2.2 Parameters for estimating noise In order to provide an objective and quantitative assessment of degree of any environmental effect it is necessary to estimate the sound level as a function of range. To estimate the sound level as a function of the distance from the source, and hence the range within which there may be an effect of the sound, it is necessary to know the level of sound generated by the source and the rate at which the sound decays as it propagates away from the source. These two parameters are: the level of sound generated by the source or Source Level (SL) and the rate at which sound from the source is attenuated as it propagates or Transmission Loss (TL) These two parameters allow the sound level at all points in the water to be specified, and in the current state of knowledge are generally best measured directly at sea, although acoustical models exist which may give reasonably reliable results for propagation from sonar systems in homogeneous deep water. However, these data have usually to be extrapolated to situations other than those in which the noise was measured; in these cases the commonest method of modelling the level is from the expression: Received Level (RL) = SL-TL Conventionally, the RL is calculated in db re 1 µpa, but a similar expression may be used to estimate the received level of other measures of sound such as its impulse. If the level of sound at which a given effect of the sound is known, an estimate may be made of the range within which there will be an effect Source level The Source Level of a source is defined as the level of sound at a nominal distance of 1m, expressed in db re 1 µpa. However, there are several assumptions implicit in this definition. Sound is made up not just of sound pressure but also a motion of the component particles of the medium (particle velocity). In the near field there are very large particle motions for a given sound pressure and this has implications for organisms sensitive to particle motion such as many fish, but less so for cetaceans. There are near-field effects around every sound source where the relationship between sound pressure and particle motion changes. In addition, some sound sources, such as airgun arrays or steerable sonars, are made up of several sound sources that are operated simultaneously. When one is within several times the diameter of the source(s) the sound field is variable. It is therefore good practice to measure the sound pressure in the far field, at 7 ICES 2004 AGISC Report

8 sufficient distance from the source that the field has "settled down", and to use this pressure to estimate the apparent level at a nominal 1m from the source. However this apparent level may not predict the actual level at ranges near an array of sources. An array of sources each of which operates at one particular source level may have a higher apparent source level far from the array. However as an animal approaches the array, the odds are low that it would experience a sound level greater than one of the individual sources from which the array is made up. A measurement of the apparent level can be made by assuming inverse dependence of pressure on the range, R, from the noise source, or by extrapolating the far field pressure. There is in general no reliable way of predicting the noise level from sources of man-made noise, and hence it is normal to measure the Source Level directly when a requirement exists to estimate farfield levels Impulsive sound Powerful impulsive sounds are generated by the use of explosives underwater, by the airgun arrays used in seismic surveying, and by some forms of construction activity such as underwater pile driving. These sources generate impulsive waves of short duration, high peak pressure, and a wide frequency bandwidth, and may consequently represent a significant hazard to underwater animals. Historically, two key parameters have been used to describe the severity of an impulsive source, the peak pressure and the impulse. The peak pressure of a blast wave P max is the maximum level of overpressure, that is, the pressure above the local ambient pressure caused by the sound. This is usually at the initial peak of the waveform and is easily read from a recording of the sound. The impulse I is defined as the integral of pressure over time and is given by I = P( t) δ t 0 where I is the impulse in Pascal-seconds (Pa.s), P(t) is the acoustic pressure in Pa of the blast wave at time t and t is time. Impulse may be thought of as the average pressure of the wave multiplied by its duration. The importance of impulse is that in many cases a wave acting for a given time will have the same effect as one of say twice the pressure acting for half the time. The impulse of both these waves would be the same. The impulse of the shock wave has been shown (Yelverton et al., 1975) to be the best predictor of damage to fish and other aquatic animals from explosives by a number of workers (Johnstone, 1985; Ross et al., 1985; Larsen and Johnsen, 1992). However, it tends to give conservative estimates for shallow water (5-10m depth) and is not considered suitable in areas having hard, reflective beds or under ice (Engelhardt et al., 1985). There are several alternative ways of measuring impulsive sounds, but given that sonars rarely generate impulsive sounds, they will not be discussed further here Sound propagation and transmission loss From the marine mammal standpoint, the most important fact about sound propagation outside the near field (near field being defined as the area where particle velocity has a greater effect than sound pressure) but comparatively close to the source is that sound pressure level drops by 6 db with each doubling of distance from the source until it encounters a boundary. A sound of 230 db one metre from the source drops to 224 at 2m, 218 at 4 m, 212 at 8m out to 190 db at 100m if the medium is homogeneous out to 100m. The reason for this extensive loss is that the energy emanating from the source expands in all directions, spread over a sphere of ever-increasing volume. This is called spherical spreading loss. Note also that the near field can be quite extensive for low-frequency sound. Losses far from the source are complex and depend on the depth of water, temperature, salinity and other factors. The second important fact about sound propagation is that not all frequencies propagate equally. High frequency sounds have a short wavelength and are absorbed by seawater and converted to heat faster than low frequency sounds that have a long wavelength. For that reason, low frequency sounds propagate farther than a high frequency sound of the same source level. However, even though they propagate further, their levels continually decrease through spreading loss. The third fact about propagation in water is that sound does not spread out equally, like light from a bulb in air. The path that sound takes through water is determined by the bottom, and by factors that change water density, mainly temperature, salinity, and depth. In summer, a sound produced near the surface over deep water tends to dive immediately toward the bottom where, because of pressure, it turns and rises, reaching the surface some 30 km away. It continues this diving and rising in a pattern called convergence zone propagation. In the winter, the same sound would ICES 2004 AGISC Report 8

9 tend to stay near the surface because propagation conditions prevent the sound from reaching great depth. If the water is very shallow, sound bounces between the surface and bottom and decreases close to shore. The relevance for marine mammals is that the exposure they receive from a human source depends not only on distance from the source and frequency, but their depth, the depth of the water, and the time of year. Sound propagation over long ranges in the ocean is therefore relatively unpredictable and also cannot be easily influenced by man. Sound may also travel horizontally through the seabed, re-emerging back into the water at a distance. Refraction and absorption further distort the impulse, leading to a complex wave arriving at a distant point which may bear little resemblance to the wave near the source. Finally, sound may be carried with little loss to great distance by being trapped by reflection and refraction between layers of water at different densities in the water column (sound channels). Sound propagating in a channel is said to have cylindrical spreading. Predicting the level of sound from a source is therefore extremely difficult, and use is generally made of simple models or empirical data based on measurements for its estimation. 2.3 Ambient noise Background, or ambient, noise occurs in all oceans and seas. There are many sources of ambient noise which may be classified as either: physical - wind driven, turbulence, seismic (earthquakes etc) and microseisms, thermal, rainfall, seabed generated and icebergs; biological - animal sounds and movement; man-made - shipboard machinery, propeller, water flow around, and discharges from, the hull. These diverse sources all contribute to the generation of background noise levels but the ambient level is not the result of noise sources alone; it also depends on propagation conditions and the absorption of sound in seawater (Francois and Garrison, 1982). Wenz (1962) and Urick (1986) describe levels of ambient noise in the ocean. The level of ambient noise in the sea increases continuously as the lower frequencies, below about 50 khz, are approached. In the northern hemisphere, from 200 Hz to 10 Hz shipping noise is dominant. In the southern hemisphere this band is less dominated by shipping. Overall trends of the level of sounds in the sea can be broken down into anthropogenic and non-anthropogenic components. For instance, there is evidence that global climate change may have resulted in higher sea states (Bacon and Carter, 1993; Graham and Diaz, 2001), which would increase ambient noise levels. Over the past few decades, however, it is likely that increases in anthropogenic noise have been more prominent. In order of importance viewed on a global scale, the anthropogenic sources most likely to have contributed to increased noise are: commercial shipping, offshore oil and gas exploration and drilling, and naval and other uses of sonar. A noise budget that covers both anthropogenic and natural sources of noise would be of considerable interest. However, no single noise budget provides a complete assessment of the potential impact of man-made sound on the marine environment. For example, a noise budget can be created that is based only on the characteristics of the source, e.g. source level, so that propagation effects are not taken into account. If, on the other hand, received sound levels are the property of interest, the relative contribution of mid frequency (1 khz - 10 khz) sources such as Navy hull-mounted sonar would be significantly reduced with respect to low frequency sources because of the ocean's selective absorption of high frequency sounds, discussed above. For example, mid frequency sonar operating at 235 db re 1 1 m in the Bahamas event was barely detectable by an acoustic range 160 km away, but airguns operating at comparable source levels were detectable over thousands of km (Nieukirk et al., 2004). In addition, since military sonars operate in a few specific areas at any given time, they probably could not be discerned in a comprehensive global noise budget in which the received sound was averaged over space and time. Therefore, local as well as global noise budgets should be constructed, particularly for critical marine mammal habitats. A further step would be to create noise budgets that take account of the auditory capabilities of various groupings of marine mammal species (odontocetes such as beaked whales, large baleen whales, etc). Ross (1987; 1993) suggested that ambient sound levels have increased by 10 db or more between 1950 and These trends are most apparent in the eastern Pacific and eastern and western Atlantic, where they are attributed to increases in commercial shipping. Ross (1993) assumed that a doubling of the number of ships explained 3 to 5 db, and greater average ship speeds, propulsion power, and propeller tip speeds explained an additional 6 db. However, more recent work (Wales and Heitmeier, 2002) calls some of these indices into question. 9 ICES 2004 AGISC Report

10 Only one actual measurement of long-term trends in ocean noise is available, and for only one site in the oceans. Andrew et al. (2002) used the same U.S. Navy acoustic array used by Wenz (1969) to make modern recordings. A low frequency noise increase of 10 db over 33 years was observed at a site off the central California coast. The explanation for a noise increase in this band is the growth in commercial shipping, in terms of both number of ships and gross tonnage. From 1972 to 1999 the total number of ships in the world s fleet increased from approximately 57,000 to 87,000, and the total gross tonnage increased from 268 to 543 million gross tons. This increase probably is not representative of the oceans as a whole because shipping density differs regionally. Mazzuca (2001) compared the results of Wenz (1969), Ross (1987), and Andrew et al. (2002) to derive an overall increase of 16 db in low-frequency noise from 1950 to This corresponds to a doubling of noise power (3 db) every decade for the past five decades, equivalent to a 7 percent annual increase in noise. During this period the number of ships in the world fleet tripled (from 30,000 to 87,000) and the gross tonnage increased by a factor of 6.5 (from 85 to 550 million gross tons) (National Research Council 2003 from McCarthy and Miller 2002). 2.4 Sonar in general Active sonar is the use of acoustic energy for locating and surveying. Sonar was the first anthropogenic sound to be deliberately introduced into the oceans on a wide scale. There are a variety of types of sonars. Sonars are used for both civilian and military purposes. They can use all sound frequencies and can be conveniently categorised into low (<1 khz), mid (1 to 10 khz) and high frequency (>10 khz). Military sonars use all frequencies, while civilian sonar uses some mid but mostly high frequencies Low-frequency sonar Low frequency sonars are used by the military for long-range (in the order of a few hundred kilometres) surveillance. The US Navy has developed the SURTASS-LFA (Surveillance Towed Array Sensor System Low Frequency Active) system that uses a vertical array of 18 projectors using the Hz frequency range. The source level of each projector is approximately 215 db re 1 1m and the ping length is 60 to 100 sec (Johnson, 2001). Over the last 10 years this system has only been used on a limited number of occasions, and is now regularly used in military testing or exercises. Many other countries of the world are developing low frequency sonar (Pengelly and Scott, 2004) to detect quiet diesel-electric submarines Mid frequency sonar Military mid frequency sonars are used to survey areas tens of kilometres in radius and are used to find and track underwater targets. A hull-mounted system (AN/SQS-53C) sonar system uses pulses in the 2 10 khz range (normally 3.5 khz) and has operated at 235 db re 1 1m with ping lengths of about 2.5 sec. A similar system (AN/SQS-56) uses this same frequency band but with lower source levels (223 db re 1 1m) (Evans and England, 2001). A very similar mid-frequency sonar is used by many navies of the world, including the Spanish navy during the Canary Islands event (see Section 4.2.4). Most usage of these systems has been confined to comparatively well-defined exercise areas, which make up a small proportion of the world s oceans. Even in these areas, activity times are relatively short and episodic, and propagation distances are small because of the frequencies involved. In addition, only a small proportion of the world s military ships carry these sonar systems. These systems were formerly used for antisubmarine work in open water, but are now most often used in coastal areas, submarine canyons or other choke points where quiet diesel-electric submarines may hide within acoustic clutter. Some non-military sonars also operate in this frequency band. Bathymetric sonars use these frequencies for wide-area, low resolution surveys. The Fugro Seafloor survey model SYS09 for instances uses both 9 and 10 khz transducers operated at 230 db re 1 1m. Sub-bottom profilers typically use 3.5 khz transducers operated at source levels of 204 db re 1 1m. The regional resolution GLORIA survey sonar uses 6-7 khz band (no source level published) High frequency sonar Military high-frequency sonars are either used in attacking (mines or torpedoes) or defending (mine countermeasure, anti-torpedo) systems and are designed to work over hundreds of metres to a few kilometres. These sonars use a wide range of modes, signal types and strengths. As with other military sonars, their usage is generally confined to exercise areas, except when they are used for commercial-like uses such as depth sounding. Fish finders and most commercial depth sounders operate at high frequencies. Usually, but not always, they project a lower power signal and have narrower beam patterns and shorter pulse lengths (a fraction of a second) than military ICES 2004 AGISC Report 10

11 sonars. These systems cannot be used at shallow depths at high powers due to cavitation (Urick, 1975). Most of the systems focus sound downwards, though some horizontal fish-finders are available. Fish finding sonars operate at frequencies typically between 24 and 200 khz, which is within the hearing frequencies of some marine mammals. Globally there are a great many recreational, fishing and commercial vessels, most of which are fitted with some sort of sonar. These vessels are most heavily used in shallow shelf-seas, with sonars used less by those merchant vessels crossing deep water areas. Usage occurs throughout the year and both by day and night. Some horizontally-acting fishfinding sonars work at frequencies at the lower end of the high-frequency range and are relatively powerful. An example is the Furuno FSV-24 sonar that operates at 24 khz and can detect and track shoals of tuna at 5000m horizontally. Source levels of these sonars are not published. Some depth finding sonars can also be powerful. Boebel et al. (2004) describe the Atlas hydrosweep DS-2 deep sea multi-beam sonar. This has source levels exceeding 220 db re 1 1m at 15.5 khz with relatively short (24ms) pulses. It is worth noting that sonars may operate at one frequency of sound, but generate other frequencies. These extraneous frequencies are rarely described and of course may have wider effects than the main frequency used, especially if the extraneous frequencies are much lower than those used (and would consequently propagate further). 3 BIOLOGICAL BACKGROUND 3.1 Hearing in cetaceans Anatomy and physiology Cetacean ear anatomy and physiology differ from the general pattern seen in terrestrial mammals in several ways. These differences are likely to be related to the specific problems of sound reception in water in contrast to air or, in odontocetes, to the echolocation abilities of the animals Sound path to middle ear Cetaceans have no outer ear and their ear canal is either vestigial (toothed whales and dolphins - odontocetes) or filled with wax (baleen whales - mysticetes). In odontocetes it is now thought that most sound enters the head and reaches the ear not through the ear canal, but through the surface of the lower jaw and is transmitted via a channel of fat to the middle ear tympanic bulla (Brill et al., 2001; Møhl et al., 1999, Norris, 1964). Anatomical and physiological features also suggest that one or more additional fatty channels, lateral to the middle and inner ear, may be involved (Ketten, 2000). There may be further pathways for sound to reach the ear. The sound path from water to middle ears in mysticetes is unknown. Anatomical similarities between bulla and middle ear ossicles in mysticetes and odontocetes, coupled with the presence of fat bodies in close contact with the middle ear (attaching onto the tympanic bulla), suggests that a fatty channel could also be involved in sound transmission in mysticetes (Thewissen, 2002) Middle ear The middle ear ossicles have undergone marked changes in cetaceans, compared to terrestrial mammals. These changes are no doubt in part or in full adaptations to underwater sound reception and connected to the loss of outer ears. The function of the middle ear is debated, but Hemilä et al. (1999) offers a model, where movement of the tympanic bulla relative to the periotic bulla is caused by sound conducted through the lower jaw fat channel and transmitted via the ossicles to the inner ear Inner ear The fundamental organisation of the inner ear of cetaceans is similar to other mammalian ears. Odontocete inner ears have anatomical specialisations for ultrasonic hearing, such as high thickness to width ratios of the basal (highfrequency) part of the basilar membrane, supplemented by additional stiffening elements along the cochlear duct (Ketten, 2000). Mysticete inner ears on the other hand have very thin and broad basilar membranes, larger than all other mammals and consistent with hearing abilities well into the infrasonic range. 11 ICES 2004 AGISC Report

12 3.1.2 Hearing in smaller odontocetes Absolute thresholds audiograms The fundamental measure of an animal s hearing ability can be represented in an audiogram, expressing the lowest sound pressures detectable by the animal in quiet conditions and at a range of frequencies. Odontocete audiograms are generally fairly similar in shape, with range of best hearing in the area khz, and best thresholds of db re. 1 µpa. The hearing thresholds of odontocetes increase slowly with ca. 20 db per decade for lower frequencies and increase steeply at high frequencies. In general, larger species seem to have an upper limit of hearing of around 100 khz, for example killer whale Orcinus orca (Szymanski et al., 1999) and false killer whale Pseudorca crassidens (Thomas et al., 1988). In contrast, smaller species have higher upper limits of hearing of around 150 khz, for example bottlenose dolphin Tursiops truncatus (Johnson, 1967) and harbour porpoise Phocoena phocoena (Andersen, 1970; Kastelein et al., 2002). As cetacean audiograms more often than not are based on only one or two individuals, one should be cautious in extrapolating especially the upper hearing limits to the species in general. A considerable natural variation between individuals of the same species may be present and it has also been demonstrated that odontocetes can suffer from age-related high frequency hearing loss (Ridgway and Carder, 1997) Dependence on duration temporal summation Some controversy exists on the question of the actual threshold determining parameter, whether it is sound pressure or sound intensity (proportional to sound pressure squared) (Finneran et al., 2002a). This question may have significant relevance when discussing damage caused by loud sounds. When it comes to discussions relating to thresholds and masking however, the ears of odontocetes behave in a similar way to other mammalian ears. For short durations, below the integration time, thresholds improve with an approximate 3 db per doubling of duration, meaning that the sound energy (intensity integrated over time) at threshold remains approximately constant. Integration time for bottlenose dolphin is between 50 and 200 ms, depending on signal frequency (Johnson, 1967). In actively echolocating odontocetes, listening for echoes of their own sonar clicks, an entirely different temporal processing seems to occur. Under these circumstances, an integration time of 265 µs has shown up repeatedly for bottlenose dolphin (Au et al., 1988; Dubrovsky, 1990; Moore et al., 1984) Masking by noise Critical bands and critical ratios have been measured for three species of odontocetes, bottlenose dolphin, beluga Delphinapterus leucas and false killer whale (Au and Moore, 1990; Johnson et al., 1989; Thomas et al., 1990). When assessing the masking effects of noise, the relevant parameter is the masking bandwidth, which provides information on the effectiveness of a given noise in masking a pure-tone signal. When masking bandwidths are calculated from critical ratios, they are roughly constant in the range of khz and around 1/12 octave in size (Richardson et al., 1995). If calculated from measurements of critical bandwidths (only available from bottlenose dolphins (Au and Moore, 1990)), which is a more direct measure of the masking interval, a value close to 1/3 octave is found, in line with values for humans and other mammals Directionality Odontocete hearing is not equally sensitive to sounds from different directions. Greatest sensitivity is for sounds coming directly towards the front of the animal, and sensitivity drops quickly as the sound source moves away from the midline. The drop is largest for higher frequencies. Threshold for a 120 khz signal is about 20 db higher 25 degrees from the midline for bottlenose dolphins (Au and Moore, 1994). The index of directionality expresses the sensitivity of the animal relative to a receptor which is equally sensitive to sounds from all directions and equal to the maximum sensitivity of the animal. For bottlenose dolphins, the index of directionality varies from 10 db at 30 khz to 20 db at 120 khz (Au and Moore, 1990). One effect of the directionality in sensitivity is a lesser influence from noise or other interfering sounds, when these sounds reach the animal from the side or from behind Hearing in larger odontocetes No audiogram or other reliable measure of larger odontocete (sperm whale Physeter macrocephalus and beaked whale) hearing is available. Carder and Ridgway (1990) obtained an audiogram using brainstem response on a sperm whale calf. In other species, this technique provides similar U-shaped responses to increasing sound frequency as behavioural techniques, but the frequency where thresholds of hearing are lowest (best) are generally much higher. Sperm whale clicks are around 5-20 khz, but Carder and Ridgway (1990) reported responses to sounds up to 60 khz. ICES 2004 AGISC Report 12

13 3.1.4 Hearing in mysticetes No audiogram or other reliable measure of mysticete hearing is available. Some inferences may be made from indirect evidence, such as the characteristics of the animals own vocalisations and morphology of their middle and inner ears. Mysticete vocalisations have fundamental frequencies from a few hundred Hz and below, to as low as Hz in blue Balaenoptera musculus and fin whales B. physalis (Edds, 1982, 1988; Watkins et al., 1987). Individual sounds may contain components up to 5-10 khz (especially grey Eschrichtius robustus and humpback whales Megaptera novaeangliae (Cerchio and Dahlheim, 2001; Crane and Lashkari, 1996). In line with observations from odontocetes and mammals in general, this suggest that the range of best hearing for mysticetes is in the similar range, i.e. from a few Hz to a few khz. This range of hearing is also supported by morphology of the basilar membrane, as described above. 3.2 Potential effects of sound on cetaceans Direct damage to hearing Potential damage to ears from underwater sound can potentially range from gross tissue damage such as that caused by the detonation of explosive charges underwater through to a temporary loss of hearing sensitivity. There is no direct evidence of tissue damage in cetaceans from underwater sound sources, but there have been no studies that have specifically investigated this. Ketten et al. (1993) found tissue damage in the ears of two humpback whales that were caught in fishing gear after explosions had occurred nearby. Exposure to high intensity noise can cause a reduction in hearing sensitivity (an upward shift in the threshold of hearing). This can be temporary (known as temporary threshold shift (TTS)), with recovery after minutes or hours, or permanent (permanent threshold shift (PTS)) with no recovery. PTS may result from chronic exposure to sound, and sounds that can cause TTS may cause PTS if the subjects are exposed to them repeatedly and for long enough. The relationship between TTS and PTS is not well-known, even for humans. However, very intense sounds can cause irreversible cellular damage and instantaneous PTS. TTS appears to be associated with metabolic exhaustion of sensory cells and anatomical changes at a cellular level. PTS may be accompanied by more dramatic anatomical changes in the cochlea including the disappearance of outer hair cell bodies and, in very severe cases, a loss of differentiation within the cochlea and degeneration of the auditory nerve. Lower frequency noises induce threshold shifts over a wider bandwidth than higher frequency noises. Finneran et al. (2002b) measured TTS in a dolphin and a beluga exposed to brief, low frequency impulses from a water gun. They compared their results with those of Schlundt et al. (2000), who measured TTS in dolphins exposed to one second tones, and those of Nachtigall et al. (2003), who measured TTS in dolphins exposed to continuous octave-band noise for 55 min. The three sets of results closely fit a 3 db per doubling of time slope. That is, if the exposure duration is doubled and the sound pressure level is reduced by 3 db (halved), the sound exposure level remains constant at about 195 db re 1 µpa 2 (s). This is an important finding because it brings some predictability to the subject of noise exposure in dolphins. There have been no direct observations of noise-induced PTS in cetaceans and such data are not likely to be obtained in the near future due to ethical concerns. However, the onset of PTS in marine mammals can be estimated by comparing the way the ear recovers from ever higher levels of TTS against similar data from terrestrial mammals that did experience PTS Non-auditory tissue damage Much research effort on the potential for anthropogenic sound to affect marine mammals has focused on auditory effects and behavioural modifications following sound exposure. Non-auditory consequences resulting from exposure to sound have historically received less attention (Crum and Mao, 1996). Studies on terrestrial mammals suggest that nonauditory tissues require exposure to sounds considerably more intense than those that affect hearing. Biologically, this extrapolation suggests that direct tissue damage can occur only very close to an intense sound source. The first hypothesis about non-auditory consequences of less intense exposures was proposed in the report of the Greek stranding event (see Section 4.2.2) and considered the concept of acoustic resonance in air spaces. All structures have a natural frequency at which they vibrate, called their resonant frequency. If such a structure is struck by an incoming sound wave of the same frequency as the resonant frequency the structure vibrates at a greater amplitude than normal; the tissues move more than normal and may tear. Acoustic resonance was suggested as a possible explanation of the Bahamian stranding (see Section 4.2.3), and the hypothesis was accepted as true by the public and the media before the 13 ICES 2004 AGISC Report

14 scientific community had adequately considered it. A workshop on acoustic resonance (Evans et al., 2002) concluded that the resonant frequencies of marine mammal lungs are too low for resonance to have been caused by mid-frequency sonar. The second hypothesized, non-auditory link between strandings and sonar exposure is acoustically mediated bubble growth (e.g. rectified diffusion) within tissues that is proposed to occur if tissues are supersaturated with dissolved nitrogen gas (Crum and Mao, 1996). Such bubble growth could result in gas emboli formation, tissue separation and increased, localised pressure in tissues, a similar scenario to decompression sickness (DCS) in human divers. Although the rectified diffusion model of Crum and Mao (1996) suggested that received sound levels of >200dB (re: 1µPa@1m) would be needed to drive significant bubble formation in marine mammal tissues, the model was run under relatively low levels of tissue nitrogen supersaturation ( %). A more recent study predicted that beaked whales, due to the typical dive profile characteristics, may accumulate over 300% nitrogen tissue supersaturation at the end of a typical dive sequence (Houser et al., 2001). This study, based on empirical observations of nitrogen tissue accumulation in bottlenose dolphins (Ridgway and Howard, 1979) and dive data from northern bottlenose whales Hyperoodon ampullatus (Hooker and Baird, 1999), suggested that beaked whales in particular may be more susceptible to acoustically mediated bubble formation than originally predicted by Crum and Mao (1996). Box 1 Decompression sickness and acoustically-mediated bubble formation Decompression sickness is the result of the supersaturation of body tissue with nitrogen and the subsequent release of bubbles of nitrogen gas. In human divers, decompression sickness is typically caused by rapid decompression following diving while using compressed air or repetitive, breath-hold dives. Unlike humans, the lungs of marine mammals collapse during a dive, limiting the nitrogen they carry to that which is absorbed into the blood stream within 60 m to 100 m of the surface, although some pinnipeds dive on expiration and lung collapse occurs at much shallower depths (e.g m in Weddell seals) (Falke et al., 1985). At greater depths, nitrogen is sequestered in non-exchanging airways. The amount of gas dissolved in specific tissues depends on dive depth, dive duration, descent and ascent rates, lipid content of the tissue, and surface intervals between successive dives. Progressive accumulation of nitrogen in tissues due to repetitive breath hold dives has been demonstrated empirically in bottlenose dolphins (Ridgway and Howard, 1979) and has been predicted to reach levels in excess of 300% supersaturation in northern bottlenose whales based on typical dive profiles (Houser et al., 2001). Although a number of anatomical, physiological, and behavioural adaptations that presumably guard against nitrogen bubble formation in marine mammals have been proposed (Ridgway 1972, 1997; Ridgway and Howard, 1979, 1982; Falke et al., 1985; Kooyman and Ponganis, 1998, Ponganis et al., 2003), it is possible that the gas emboli and associated lesions found in cetaceans in the Canary Islands and in the UK (Jepson et al., 2003; Fernandez et al., 2004, in press) could be caused by disruption of these evolutionary adaptations to deep diving. Anatomical and physiological adaptations to diving are unlikely to alter in the short course of acoustic exposure, but behavioural changes in response to sonar might. For example, in experiments northern right whales Eubalaena glacialis responded to novel acoustic stimuli by a combination of accelerated ascent rates and extended surface intervals at received sound levels as low as 133dB re 1µPa@1m (Nowacek et al., 2004). If beaked whales respond similarly they could experience excessive nitrogen tissue supersaturation driving potentially damaging bubble formation in tissues via a similar mechanism to the human diver that incurs DCS due to too rapid an ascent. Alternatively, physical mechanisms (e.g. rectified diffusion) exist for acoustically-mediated bubble formation in tissues already supersaturated with nitrogen (Crum and Mao, 1996; Houser et al. 2001). It is therefore theoretically possible that sonar transmissions (of low, mid or high frequency) could directly initiate or enhance bubble growth in tissues were sufficiently supersaturated with nitrogen and if the received sound pressure levels were of sufficient intensity. However, there is as yet no scientific evidence for any of the steps in these postulated chains of events. A (US) Marine Mammal Commission Workshop on beaked whales and anthropogenic noise considered it important to test the bubble hypothesis, and prioritised a programme of research that incorporates both acoustically mediated bubble formation and bubble formation via a DCS-like mechanism, and includes the use of controlled exposure experiments (Cox et al., in prep.). Even more recently, the first evidence of gas and fat emboli and acute and chronic gas bubble lesions has been reported in a number of cetacean species stranded in Europe. In the UK, ten stranded cetaceans comprising four Risso s dolphins, four common dolphins, a Blainville s beaked whale and a harbour porpoise had acute and chronic lesions in liver, kidney and lymphoid tissue (lymph nodes and spleen) associated with (predominantly) intravascular gas bubbles (emboli) (Jepson et al., 2003, in press; Fernández et al., 2004). These animals stranded singly and the etiology of these lesions (including whether or not they were exposed to any form of acoustic activity) is unknown. However, a suite of widely disseminated microvascular haemorrhages associated with gas and fat emboli, lesions highly consistent with DCS, were found in ten beaked whales that died as part of a mass stranding of 14 beaked whales in the Canary Islands linked to an international naval exercise (Neo Tapon) in September 2002 (see Section 4.2.4). The Canaries findings are important for understanding effects on tissues as they are the first to be based on fresh material. In other similar ICES 2004 AGISC Report 14

15 incidents, either tissues were not examined, or were examined much later. The gas bubble hypothesis is relatively new and has received much recent theoretical attention and evidence, however there has as yet been little scientific testing of it. Such testing is needed and necessary before a full judgement of the hypothesis can be made Masking and changes in vocal behaviour Cetaceans use sound for a number of purposes including communication, searching for food and detecting predators. In all cases, a cetacean needs to hear a sound, either originating from itself (with an echo reflected from a target) or originating somewhere else and may not be very loud. In order to detect the sound, the sound has to be louder than (or be able to be differentiated from) the ambient sound level. The hearing mechanisms or auditory processing of the whale also has to be sensitive enough to detect this difference. An increase in ambient noise could have a number of effects depending on the particular use of sound by the cetacean. If a sound is propagating with cylindrical spreading, a 10 db increase in ambient sound could effectively reduce the maximum range at which a sound can be heard to a tenth of the original range (a 20 db increase could reduce this to a hundredth of the range). It is likely that cetaceans can compensate to some extent to this increase in ambient noise. Belugas adjust their echolocation clicks to higher frequencies and higher source levels in the presence of increased background noise (Au et al., 1985). Long-finned pilot whales Globicephala melaena changed the type of vocalisation in the presence of military sonar signals (Rendell and Gordon, 1999). Belugas have been observed to increase call rates and shift to higher call frequencies in response to boat noise (Lesage et al., 1999). Some humpback whales lengthened their song cycles when exposed to the LFA source (Miller et al., 2000; Fristrup et al., 2003); increasing the redundancy of the song may improve communication in a noisier channel. Note that it is difficult to separate the two possible causes of these changes masking by the sound and direct behavioural response to the sound. Thus, one general effect of the increase in ambient ocean noise could be to affect several vocal characteristics or behaviours of cetaceans. The degree to which these changes significantly affect the animals is not known and will be case dependent. Sonar is a lesser contributor to the overall ocean noise budget than other sources of anthropogenic sound. Surveys conducted by NOAA Fisheries showed that during the last 20 of the 33 years covered by the Andrew et al. (2002) study all cetaceans in west coast of the USA increased by an average of 8.2%. A mark-recapture study on humpback whales showed a 7.2% increase in this same area, which bolster s the survey estimates (Barlow, 1994; NOAA, unpublished data). The increases noted in section 2.3 in ambient noise level involved frequencies used for communication by blue, fin, and grey whales. If ambient noise has untoward effects on mysticete populations through masking, it is not apparent at the levels reported by Andrew et al. (2002) Behavioural reactions Besides the changes in vocal behaviour outlined above, many possible changes in behaviour could occur in the presence of additional noise. Behavioural responses may range from changes in surfacing rates and breathing patterns to active avoidance or escape from the region of highest sound levels. Several studies suggest that bowhead whales Balaena mysticetus follow a pattern of shorter surfacings, shorter dives, fewer blows per surfacing, and longer intervals between blows when exposed to anthropogenic noise, even at moderate received levels (114 db re 1µPa). Many now hypothesise that the mechanism(s) underpinning the phenomenon of beaked whale mass strandings linked to naval sonar are initially triggered by a behavioural response to acoustic exposure rather than a direct physical effect of acoustic exposure (Jepson et al., 2003; Fernández et al., 2004, in press; Cox et al., in prep.) (see section 3.2.2). The first potential pathway entails a simple behavioural response to sound that leads directly to stranding, such as swimming away from a sound into shallow water. An alternative scenario involves a behavioural response leading to tissue damage. Such responses may include a rapid ascent, staying at depth, or remaining at the surface, and these could lead to gas bubble formation, hypoxia, hyperthermia, cardiac arrhythmia, hypertensive hemorrhage, or other forms of trauma (Cox et al., in prep.). Of these, the hypothesis that a behavioural change could lead to gas bubble formation via a mechanism similar to decompression sickness (Jepson et al., 2003; Fernández et al., 2004, in press) is seen as a priority for future research (Cox et al., in prep.). Beaked whales might also experience tissue damage directly from sound exposure, such as acoustically mediated bubble formation and growth (Crum and Mao, 1996), vestibular response that leads to stranding, acoustic resonance, or hemorrhagic diathesis, and all of these could lead to behavioural alterations in beaked whales, stranding and death (Cox et al., in prep.). Marine mammal responses also appear to be affected by the location, motion, and type of onset of a sound source. Bowheads are more responsive to overflights of aircraft when they are in shallow water (Richardson and Malme, 1993). Fin whales are more tolerant of a stationary than a moving source (Watkins, 1986). Humpback whales are less likely to 15 ICES 2004 AGISC Report

16 react to a continuous source than to one with a sudden onset (Malme et al., 1985). In the St. Lawrence River, belugas are less likely to change their swimming and diving patterns in the presence of vessels moving at low speed than in the presence of fast-moving boats (Blane and Jaakson, 1994). In Alaska, belugas feeding on river salmon may stop and move downstream in response to noise from small boats, whereas they are relatively unresponsive to noise from fishing boats (Stewart et al., 1982). In Bristol Bay, belugas continue to feed even when surrounded by fishing vessels, and they may resist dispersal even when purposely harassed (Fish and Vania, 1971). This context-dependent response to sound and disturbance illustrates the difficulty of extrapolating results from captive animals to those in the wild. Few studies have been designed to document long-term responses to anthropogenic noise by marine mammals. At Guerrero Negro Lagoon in Baja California, Mexico, shipping and dredging noise associated with a salt works may have induced grey whales to abandon the area through most of the 1960s (Bryant et al., 1984). After ship traffic declined, the lagoon was reoccupied, first by single whales and later by cow-calf pairs. Killer whales in the British Columbia region were displaced from Broughton Archipelago during , a period when acoustic harassment devices were in use at existing salmon farms (Morton and Symonds, 2002). Displacement by sound from areas could have effects on individual animals and populations, probably depending on the distance and persistence of the displacement. These effects may not become immediately apparent and could be modified by habituation, sensitisation, hearing loss, physiological damage and stress. Noise would be biologically significant if it induced long-term abandonment of an area important for feeding, breeding or rearing the young, as it may lead to reduced fecundity, carrying capacity, or both. Social disruption brought about by noise may be especially important if mother/calf pairs become separated. The individual and population level effects of non-lethal disturbance are also likely to be dependent on body size and life history. Generally, larger whales need to balance their energy budgets over time-spans of months (up to a year) (Boyd, 2002). Many larger whales migrate to high latitudes to feed during the summer, storing energy in the form of blubber. These whales return to breed in warmer, less productive tropical waters during which time they may fast and rely on their blubber for energy. Smaller whales and dolphins are likely to balance their budgets over shorter (days to a month) periods. Smaller cetaceans may be more susceptible to shorter term disruption of foraging. The consequences of disturbance are thus time and space specific for each species. 4 CETACEANS AND SONAR 4.1 Marine mammals There are globally around 120 species of marine mammal (the precise figure depending on the taxonomy used). Since ships of European countries operate globally, all of these species could potentially be affected by sonar from ships of European countries. In the waters of Member States of the European Union waters, there are about 30 species of cetacean and 10 species of seal. Each species of marine mammal has a unique geographical distribution though, with some distributions being better known than others are. Thus which species is affected by which sonar usage depends heavily on the location in which the sonar is used (and the local propagation characteristics of the particular frequency band being used by the sonar). Geographic distributions of cetaceans in north-west European waters have been described by Reid et al. (2003), while preliminary maps for Mediterranean Sea were assembled by Beaubrun et al (1995). The harbour porpoise is the only resident cetacean species in the Baltic, and here the species is rare and confined mostly to the south and west of the sea (ICES, 2003). The distribution of cetaceans in the Bay of Biscay, off Iberia and around the Macronesian Islands has not been mapped systematically. Although many species may live in a given area, sonar-related strandings and deaths typically involve mostly the beaked whales present. Few, if any, effects of sonar on other species have been observed in European waters. Consequently, rather than review the knowledge of distribution of all marine mammals in EU waters here, we focus on knowledge (and lack thereof) of beaked whale distribution and refer readers to the sources mentioned above for other species. 4.2 Beaked whales There are some 20 species of beaked whale known globally at present, but given that two of these have only been discovered and described in the past twenty years, it would not be surprising if further species were found. One reason for this lack of taxonomic certainty is that all beaked whales appear to live in the deep ocean or on the margins of the continental shelves. All species appear capable of diving to great depth, staying underwater for many minutes (more than an hour in some cases) and then only being at the surface for a relatively short time before diving again. Surfacing behaviour is frequently relatively inconspicuous. Many of these species are shy and respond to the presence of ships by prolonged diving. Not surprisingly, this group of comparatively large mammals is one of the least known on the planet and much remains to be discovered about all aspects of their biology. Six/seven species have been recorded in European waters (Table 4.2.1), but many sightings of the group are not identified to species level. ICES 2004 AGISC Report 16

17 Table Beaked whales recorded in European waters Cuvier s beaked whale Ziphius cavirostris Northern bottlenose whale Hyperoodon ampullatus Sowerby s beaked whale Mesoplodon bidens (Gray s beaked whale Mesoplodon grayi one stranding record (1927) only) True s beaked whale Mesoplodon mirus Gervais beaked whale Mesoplodon europaeus Blainville s beaked whale Mesoplodon densirostris Most records of beaked whales in European waters come from deep waters at or beyond the shelf break. Reid et al. (2003) mapped a group of sightings to the north and west of Scotland (Figure 4.2.1), but this was one of the few areas where considerable survey effort had been expended in deeper water by those contributing to their atlas. It was however noticeable that the deeper cold-water area north of the Wyville Thomson ridge appeared to be less important than the area south of the ridge. Northern bottlenose whales were less tied to the areas near shelf breaks, but occurred throughout deeper water (Figure 4.2.2) in the survey reported by Reid et al. (2003). Further south, surveys from ferries running from the UK across the Bay of Biscay have shown that the deep-water area and the surrounding shelf slope/canyons are important for Cuvier s beaked whale (Cresswell and Walker, 2001; Coles et al., 2003), and other beaked whales have been recorded in this deep water area, but not in shallower waters. Cuvier s beaked whale is the only beaked whale regularly present in the Mediterranean Sea (Notarbartolo di Sciara, 2002). Beaubrun et al. (1995) records seven sightings of live Cuvier s beaked whale all from shelf break areas. The depressions and the deep trenches that surround the coasts of Greece appear to be good habitat for this species. Strandings and sightings during the last decade show the importance of Greek Seas for this species (Frantzis et al., 2003). The yearly average of Cuvier s beaked whale strandings in Greece was 5.6 individuals (C.L.95%=2.78) for the decade (the mass stranding of May 1996 (see Section below) excluded). Although underestimated (because no official stranding network was established in Greece before 1992), this number is significantly higher than the respective average for each of the three northern countries of the west and central Mediterranean (Spain 1.9, France 0.2, Italy 2.6), and higher (although not significantly) than their sum. Most of the strandings that occurred in Greece were recorded along the Hellenic Trench, which runs all around the west and south Greece, and marks the limits of the European continental shelf. Surveys conducted along the Hellenic Trench confirmed that Cuvier s beaked whales are abundant (Frantzis et al., 2003). It is reasonable to assume that all suitable shelf break areas in European Seas are likely to form habitat for beaked whales, with some distributions extending to deeper water. It is not known how resident any individual or group of beaked whale might be to any particular area off Europe. 17 ICES 2004 AGISC Report

18 Figure Distribution of effort-based sightings of beaked whales recorded by Reid et al. (2003). The background grey shading indicates hours of searching for each quarter ICES rectangle. The 200 and 500m depth contours are shown, indicating the area of the continental shelf break. Red circular dots indicate sightings rates per hour of searching. Figure Distribution of effort-based sightings of northern bottlenose whales recorded by Reid et al. (2003). See Figure for key to symbols used Review of literature on effects of sonar on beaked whales Cuvier s beaked whale is a deep-diving, pelagic cetacean that until recently was believed to rarely mass-strand (Heyning, 1989). Only seven strandings of more than four individuals were recorded by Frantzis (1998) from 1963 to ICES 2004 AGISC Report 18

19 1996 worldwide, but more incidents in this period have come recently to light. On most of these occasions, mass strandings showed atypical characteristics unlike those that occur with other whales. This suggested that the cause had a large synchronous spatial extent and a sudden onset. Such characteristics are shown by sound in the ocean. Cetaceans and particularly the deep-diving whales were known to be especially affected by low and mid-frequency anthropogenic sound, even at quite low received levels (Watkins et al., 1985; Finley et al., 1990; Finley and Greene, 1993; Bowles et al., 1994; Richardson and Würsig, 1997). Research on LFAS began by NATO in 1981 (NATO-Saclantcen, 1993) and the US Navy s research on SURTASS LFA began about 1986 and a statement on its environmental impact was formally initiated in July It is worth noting that the first atypical mass stranding of Cuvier s beaked whale was in 1963 (Tortonese, 1963), shortly after the time that a new generation of powerful mid frequency tactical sonars became widely deployed (Balcomb and Claridge, 2001). Hildebrand (2004) published a list (compiled by James Mead) of strandings of two or more Cuvier s beaked whales based on records at the Smithsonian Institution and recent literature (Table ). This list is unlikely to be complete, but it represents all cases presently known. In only four of the cases, Greece 1996, Bahamas 2000, Madeira 2000 and Canary Islands 2002, is it documented that navy vessels were in the area, operating sonar at the time and place of the stranding, and partial or complete necropsies were undertaken. No necropsy results are available for any of the other events. It should be noted that it has proven very difficult to demonstrate whether or not military sonar was in use sufficiently near the stranding sites to be considered as a possible cause of the stranding. It is recommended that in future, systematic efforts be made to determine if any abnormal noise has been made near mass strandings. It is worth noting also that some other strandings, not categorised as mass strandings, could be caused by the same mechanism as behind the mass stranding. These records have not been reviewed for possible correlation with presence of naval vessels. Since the stranding in the Kyparissiakos Gulf, Greece in 1996 (see Section below), there has been increasing attention paid to the effects of sonar. Sections illustrate some of the findings based on three case studies of incidents. Table Strandings involving at least 2 Cuvier s beaked whale Ziphius cavirostris (Zc) (after Hildebrand, 2004; Brownell et al., 2004; Martín et al., 2004; Litardi et al., 2004). Strandings refers to individuals that became stranded on beaches and does not imply death (some were redirected out to sea and their fate is unknown). Items listed U.S. Fleet? and Naval manoeuvres represent mostly the word of locals that military ships might have been in the general area and cannot be taken as necessarily linked. These records also represent the only known multiple stranding events for Gervais beaked whale Mesoplodon europaeus (Me) and Blainville s beaked whale Mesoplodon densirostris (Md). Year Location Species (numbers) Correlated activity, when available 1914 New York, United States Zc (2) 1963 Gulf of Genoa, Italy Zc (15+) Naval manoeuvres 1963 Gulf of Genoa, Italy Zc (15+) Naval manoeuvres 1963 Sagami Bay, Japan Zc (8-10)?US Fleet 1965 Puerto Rico Zc (5) 1966 Ligurian Sea, Italy Zc (3) Naval manoeuvres 1968 Bahamas Zc (4) 1974 Corsica Zc (3), Striped dolphin (1) Naval patrol (?not sonar) 1974 Lesser Antilles Zc (4) Naval explosion 1975 Lesser Antilles Zc (3) 1978 Sagami Bay, Japan Zc (9)?US Fleet 1978 Sagami Bay, Japan Zc (4)?US Fleet 1979 Sagami Bay, Japan Zc (13)?US Fleet 1980 Bahamas Zc (3) 1981 Bermuda Zc (4) 1981 Alaska, United States Zc (2) 1983 Galapagos Zc (6) 19 ICES 2004 AGISC Report

20 1985 Canary Islands Zc (12+), Me (1) Naval manoeuvres 1986 Canary Islands Zc (5), Me (1), Ziphiid sp. (1) 1987 Canary Islands Me (3) 1987 Italy Zc (2) 1987 Canary Islands Zc (2) 1988 Canary Islands Zc (3), bottlenose whale (1), pygmy sperm whale (2) Naval manoeuvres 1989 Sagami Bay, Japan Zc (3)?US Fleet 1989 Canary Islands Zc (15+), Me (3), Md (2) Naval manoeuvres 1990 Sagami Bay, Japan Zc (6)?US Fleet 1991 Canary Islands Zc (2) Naval manoeuvres 1991 Lesser Antilles Zc (4) 1993 Taiwan Zc (2) 1994 Taiwan Zc (2) 1996 Greece Zc (12) LFAS trials (see Section 4.2.2) 1997 Greece Zc (3) 1997 Greece Zc (9+) Naval manoeuvres 1998 Puerto Rico Zc (5) 2000 Bahamas Zc (8), Md (3), Ziphiid sp. (2), minke whale (1), Balaenoptera sp. (2), Atlantic spotted dolphin (1) 2000 Galapagos Zc (3) Naval mid-frequency sonar (see Section 4.2.3) 2000 Madeira Zc (3) Naval mid-frequency sonar 2001 Solomon Islands Zc (2) 2002 Canary Islands Zc, Me, Md (15-17 whales) Naval mid-frequency sonar (see Section 4.2.4) 2002 Mexico Zc (2) RV Ewing seismic Møhl (2004) points out that sperm whale clicks bear some resemblance to those of tactical sonars (Table ). The main differences are the ping energy a receiving animal would need to be times closer to a sperm whale than a tactical sonar to receive the same sonic energy. The duty cycle (or proportion of overall time that the noise is made) is also much higher in tactical sonar and the directionality of each is different sperm whales emitting a very narrow beam of sonic energy compared to the wide radiation pattern of the tactical sonars. Møhl (2004) felt that this similarity in properties between a natural noise source compared with the novel sources indicated that behavioural rather than physiological causes would be more likely to cause the multiple beaked whale strandings. Table Properties of sonar signals from sperm whales and tactical sonars (Møhl, 2004). Sperm whale AN/SQS-56 AN/SQS-53C Source level (db re 1 µpa) Ping duration (ms) Ping energy (db re 1 µpa*s) Repetition rate (pings/s) Duty cycle (%) Frequency (khz) , 7.5, , 3.3 Spectrum type Broad band Narrow band Narrow band Directionality (half-power, 4 360x30 120x40 half angle, degrees) ICES 2004 AGISC Report 20

21 4.2.2 Case study: Greece During the early hours of the morning of 12 May 1996, Cuvier s beaked whales started to strand alive in several locations along Kyparissiakos Gulf (a long sandy beach alongside the Hellenic Trench in the west coast of the Peloponnese (Frantzis, 1998, Figure ). The strandings continued until the afternoon of 13 May A few more specimens (4-5) were reported as stranded and rescued, entangled and rescued, or swimming very close to the coasts during the next 3 days, however, only one of these reports could be confirmed. In total, 12 stranded whales were recorded on 12 and 13 May. They were spread along 38.2 kilometres of coast and were separated by a mean distance of 3.5 km (s.d. = 2.8, n = 11) (Fig ). Another whale stranded on 16 May and was driven back to the open sea. Two weeks later, one more animal was found decomposing on a remote beach of the neighbouring Zakynthos Island, 57 km away from the closest stranding on the mainland. Eleven of the whales were measured and sampled. Nine of them were immature males with no erupted teeth and two were females. The recorded spread of the stranded animals in location and time was atypical, as whales usually mass-strand at the same place and at the same time. The term atypical mass stranding has been proposed for the recorded strandings as opposite to typical mass strandings known mainly from pilot whales Globicephala sp. and false killer whales (Geraci and Lounsbury, 1993). Figure Position, sex and total length of the 14 Cuvier s beaked whales that were recorded during, or shortly after the mass stranding of 12 and 13 May 1996 in Kyparissiakos Gulf, Greece. Necropsies of eight stranded animals were carried out, but no apparent abnormalities or wounds were found. These necropsies were limited to basic external examination and sampling of stomach contents, blood and skin. No ears were collected; no entire organs or histological samples were conserved because of many problems related to permits, lack of facilities and means, and lack of relevant knowledge and trained specialists. Stomach contents had variable quantities of squid remains (like beaks and ocular lenses) from three different squid species. Many of them contained cephalopod flesh, indicating that recent feeding had taken place. All available information regarding the conditions associated with the mass stranding of May 1996 was gathered, and many potential causes were listed and examined. The most important of them were major pollution events, important tectonic activity, unusual geochemical/physical/meteorological events, magnetic anomalies in the area, epizootics and conventional military exercises. However, none of the potential causes listed above coincided in time with the mass stranding or could explain its characteristics (NATO-Saclantcen, 1998). Several months after the mass stranding a warning to mariners issued by the Greek Hydrographic Service was found by cetacean researchers that provided significant relevant information. This warning (586 of 1996) stated that sound-detecting system trials were being performed by the NATO research vessel Alliance from 24:00 11 May to 24:00 15 May - a period that encompassed the mass stranding. The officially declared area where the sea trials had been carried out enclosed all the co-ordinates of the stranding points. The tests performed were for Low Frequency Active Sonar (LFAS; term used by NATO to describe their dual low- and mid-frequency active sonar), a system that introduces very high level of low and medium frequency sound into the marine environment to detect quiet diesel and nuclear submarines. Detailed information regarding the time schedule, the runs (Figure ) and the specific sound characteristics of the transmissions became declassified and available through NATO-Saclantcen by the autumn of 1998 (NATO-Saclantcen, 1998). The Alliance was using high power active sonar, transmitting simultaneously to both low ( Hz) and mid ( khz) frequencies, at a maximum output of 228 db re 1 1 m, which enables long detection ranges. 21 ICES 2004 AGISC Report

22 Figure Routes of the first and second day of the LFAS tests (12 and 13 May 1996) according to NATO- Saclantcen (1998). The black arrows indicate the stranding positions of the whales during the same days, with the position of the whale found in Zakynthos Island. Although the available data in 1996 could not directly prove that the use of active sonars caused the mass stranding in Kyparissiakos Gulf, the evidence clearly pointed to the LFAS tests. The main arguments and the supporting evidence are listed below: At least 12 of the 14 animals stranded alive in an atypical way. The condition of the stranded animals, along with the analyses of their stomach contents was not consistent with pathogenic causes (which anyway are not known to provoke atypical mass strandings). No unusual environmental events occurred before or during the stranding (e.g. tectonic activity, magnetic anomalies, geophysical or geochemical events, meteorological events etc.). The stranding characteristics suggested a cause with large synchronous spatial extent and sudden onset (i.e. those shown by sound in the ocean). Most importantly, the probability for the two events (i.e. the LFAS tests and the mass stranding) to coincide in time and location, while being independent, was extremely low. In other words if the 16.5-year period before the mass stranding is considered (1981 was chosen arbitrarily because this was the year that NATO started to experiment on LFAS, and we are sure that no mass stranding, nor other tests of LFAS had occurred in the area since that year), the probability of a mass stranding occurring for other reasons during the period of the LFAS tests (i.e. from 12 to 15 May 1996 instead of any other day) is less than 0.07% Today, after three repeated mass strandings that followed the Greek case with similar characteristics and always in close association with naval exercises and use of mid-frequency active sonar in the Bahamas (see below), Madeira, and Canary Islands (see below), there is no dispute in the scientific community regarding the cause of the mass stranding in Kyparissiakos Gulf. If, after this case, more effort had been invested in mitigation, and military sonar had not been used in sea areas known to have many beaked whales, then the mass strandings that followed may have been avoided. The Cuvier s beaked whale stranding history of Kyparissiakos Gulf (Fig ) shows that although no mass strandings had been recorded before the 12 May 1996 the strandings of this species were not rare. The average stranding rate was 0.88 individual/half year (s.d. = 0.99, n = 8). After the mass stranding of May 1996, the stranding rate was reduced to less than one third of what it was before the mass stranding (0.25 individual/half year, s.d. = 0.45, n = 12). This alarming result indicates that the damage could be significantly higher than the death of the stranded whales. Many others may have left the area or may have died in the deep offshore waters. ICES 2004 AGISC Report 22

23 Fig : The stranding history of Cuvier s beaked whales in Kyparissiakos Gulf from 1992 to Case study: Bahamas Introduction On March 14 and 15, 2000, five U.S. Naval ships using mid frequency (2-10 khz) sonar transited the Northeast and Northwest Providence Channels of the Bahamas Islands in an anti submarine warfare exercise lasting 16 hours. The ships were using two types of mid frequency sonar, designated AN/SQS-56 and AN/SQS-53C, that differed somewhat in their operating characteristics. The AN/SQS-56 closely resembles mid-frequency tactical sonars used by many other navies of the world. The ships operated in two loosely coordinated groups that passed through the channel six hours apart. Beginning on March 15 only hours after the first group of ships passed, and continuing for the next 36 hours, 17 cetaceans were found stranded dead or alive, or in shallow water, along a 240 km stretch of the Northeast and Northwest Providence Channels on three islands (Figure ). The Bahamas Marine Mammal Survey discovered the stranding, and they, Dr. Alan Bater, veterinarian for Bahamas Department of Fisheries, and members of the public pushed some of the stranded animals back into deeper water, and preserved for post mortem examinations tissues from those that died. The preserved specimens were shipped to the U.S. mainland and distributed among several pathologists for broad-based analysis of the cause of death. The U.S. Navy was informed about the event and immediately started summarizing ship tracks and times, and modelling acoustic propagation from the sonars. The National Oceanic and Atmospheric Administration (NOAA) sent representatives of its stranding program to the Bahamas to assist in handling the biological specimens. It later sent specimens to a number of researchers for histological and toxin studies. Navy and NOAA each prepared verbal reports of their own findings, and in June 2000, the two agencies met for the first time and exchanged information. Subsequently, each agency prepared a written version of its report and submitted them to two editors (Cdr. Paul Stewart for Navy and Dr. Roger Gentry for NOAA) who compiled an interim report on progress to date. The report included the results of NOAA acoustic monitoring of the Bahamas region on the days of the sonar exercise and stranding (Evans and England, 2001) Findings The Navy sonar systems produced a sound approximately every 12 seconds. Except for one four hour period when one of the ships produced source levels that are classified, the source levels of all ships during the remainder of the exercise did not exceed 235 db re 1 µpa. Complex propagation modelling showed that because of a surface duct, the sound was largely confined to the top 200 m of the water column, and that in many areas of the channel levels of 160 db re 1 µpa would have occurred. Reverberation from the walls or floor of submarine canyons is not thought to have added much to these levels because of the surface duct. Whale locations at sea were unknown, so received levels cannot be estimated with confidence. The animals that stranded included Cuvier s and Blainville s beaked whales, minke whales Balaenoptera acutorostrata, and an Atlantic spotted dolphin Stenella frontalis. An animation that plotted ship positions by time, and the time and place of each stranding showed a close temporal and spatial correlation for all but the spotted dolphin. Seven of the stranded animals died, including five Cuvier s beaked whale, one Blainville s beaked whale, and a spotted dolphin. The latter may have died of causes not associated with acoustic exposure, and in a very different location than 23 ICES 2004 AGISC Report

24 the beaked whales. Four of the beaked whales showed some evidence of auditory structural damage, including bloody effusions near and around the ears. The two freshest specimens showed subarachnoid haemorrhage and blood clots in the lateral brain ventricles. It is reasonable to assume the haemorrhages were acoustically induced. The immediate cause of death appeared to be cardiovascular collapse and physiological shock which together commonly result in death after stranding. NOAA s investigation considered every possible cause of the stranding event, and eliminated all except sonar as the triggering event. Explosions were eliminated by NOAA s acoustic recordings. The evidence that most strongly suggested sonar as the triggering event was the close temporal and spatial match between sonar passage and the stranding events. The underlying mechanism by which sonar had this effect is still not known. It is possible but highly doubtful that direct acoustic exposure of tissues caused the lesions observed. All animals would have had to be very close to the vessels to receive such exposure, which seems unlikely. It is possible that sonar triggered some kind of unfavourable behavioural response which led to stranding and to subsequent tissue injury. It is also possible that some injury occurred before and some after stranding. Figure Locations of seventeen marine mammals that were stranded following anti-submarine exercises in the Northeast and Northwest Providence Channels, Bahamas Islands, on March Initials indicate scientific names, numbers show specimen number. Zc = Cuvier s beaked whale Ziphius cavirostris, Md = Blainville s beaked whale Mesoplodon densirostris, Sf = Atlantic spotted dolphin Stenella frontalis, U = unidentified ziiphid, Ba = unidentified baleen whale Conclusions The association of mid frequency sonar with this, the Madeira, and the Canary Islands strandings suggests that it was not the low frequency component of the NATO sonar that triggered the stranding in Greece in 1996, but rather the mid frequency component. NOAA is arranging for a final report of the Bahamas event to be written after team members become more familiar with Canary Islands material and after they revisit all of the analyses that went into the Interim report. ICES 2004 AGISC Report 24